metadata
dataset_info:
features:
- name: conversations
list:
- name: from
dtype: string
- name: value
dtype: string
- name: state
sequence: string
- name: tokens
dtype: int64
- name: class
dtype: string
splits:
- name: train
num_bytes: 4264463317
num_examples: 3585625
download_size: 1245131355
dataset_size: 4264463317
configs:
- config_name: default
data_files:
- split: train
path: data/train-*